Professional Documents
Culture Documents
JEP2006 McCleary&Barro - Religion and Economy
JEP2006 McCleary&Barro - Religion and Economy
in response to advances in education and science and to movements away from the
vicissitudes of agriculture and toward the greater economic security of advanced,
urbanized economies. In Marx's (1859 [1913]) analysis, the decline of religion is
one manifestation of a broader trend toward "modernization."
affect the economy by fostering traits such as work ethic, honesty (and hence trust),
thrift, charity, hospitality to strangers and so on. By enhancing these traits, greater
religiosity could spur investment and economic growth. Wesley's (1760) views, cited
by Weber, are similar in some respects. Wesley famously urged his congregants to
"gain all you can, save all you can, give all you can." However, he regretted that he
had been more successful in promoting the first two tenets than the third. But the
first two-akin to Weber's work ethic and thrift-are probably more important
than charity as underpinnings of a productive economy. Wesley also regretted that,
as his congregants became richer, they became less devout-thus giving an early
empirical expression of the secularization hypothesis.
A key point about religion in the Weberian framework is that religious beliefs
are what matter for economic outcomes. This approach contrasts with a social-
capital/cultural perspective, in which the networking associated with attendance at
formal religious services could be what promotes growth. This alternative perspec-
tive trivializes religion by viewing participation in formal religion as just one of
many ways to build social capital or to form a communal culture. For Weber, houses
of worship were not merely forms of social clubs. The special feature of religion is
its potential influence on beliefs that reinforce particular traits and values.
We take the position that religion is sui generis. Otherworldly compensators
like salvation, damnation and nirvana are great motivators of behavior in this world.
Thus, beliefs in these compensators can raise productivity by fostering individual
traits such as honesty, work ethic and thrift. In other contexts, the powerful force
from afterlife beliefs can promote anti-social actions, such as violence-the so-
called "dark side of religion." In either context, the social capital and cultural
aspects of religion-communal services, rituals, religious schools-are significant
only to the extent that they influence beliefs and, hence, behavior. In fact, for given
beliefs, more time spent on communal activities would tend to be an economic
drag, at least as measured by market output (GDP). Moreover, the costs of formal
religion include not only the time of congregants and religious officials but also the
resources expended on buildings, precious objects and so on. Our general view is
that believing relative to belonging (or attending) is the main channel through
which religion matters for economic and other outcomes.
enment and knowledge. Catholicism, Hinduism, and Islam have medium salvific merit, in
that individuals have some but not necessarily decisive influence on salvation.
Each of the major religions has some mechanism for promoting work effort and
wealth accumulation, which contribute to economic success.2 However, the incentive
to acquire and accumulate property is limited in Buddhism, because the sharing of
wealth tends to be stressed. One reason for this emphasis is to ensure the survival of the
community. By spiritually rewarding networks of mutual aid and charitable acts,
religion lowers the uncertainties of daily life. That is, charity is a form of communal
insurance, which can be efficient if the society has a lot of uncertainty, such as that
associated with agriculture. Private charity supported by religion would be particularly
useful if the society lacks formal structures, such as insurance markets and government
welfare programs, to deal with individual uncertainties (Gill and Lundsgaarde, 2004;
Scheve and Stasavage, 2005). Buddhism also helps to ensure its own survival by linking
salvific merit to particular acts-giving financial aid to the religious class, praying
communally, and constructing religious edifices.
Although Hinduism and Buddhism do not have heaven and hell in the
Judeo-Christian sense, the believers who perform their obligations are effectively
reincarnated into heavenly intermediate stages. Those who fail to perform their
obligations are reincarnated into intermediate, transitory stages of purgatory.
Performing more than is required can also help a person move into a higher stage
of heaven by earning salvific merit and can shorten a person's stay in purgatory
(Keyes, 1983, p. 267; Conze, 1963, p. 41). Reincarnation is a process that enables a
person to get rid of bad karma and move toward a higher state of perfection or
enlightenment. In short, Hinduism and Buddhism are belief systems about how to
attain perfection, which can be interpreted as a form of salvation.
At the other end, Calvinist Protestantism, with its emphasis on predestination,
seems at first glance to be weak on economic incentives. After all, according to
Calvin (1585, pp. 237-239), a person is either one of the elect or not, and no good
works or other worldly acts can do anything about it. However, the uncertainty
about salvation is also stressed, as is the motivation to gain some kind of sign that
one has been chosen. The Reformed churches, which closely follow Calvin's
theology, stress outward or social signs of salvation. Human industry and thrift that
result in material success are the clearest possible signals that God has chosen the
person as one of the elect who will be saved (Calvin, 1584, pp. 194-196). Economic
success is therefore highly valued, but charitable acts are downplayed, if not
condemned, as going against God's will-for example, by promoting idleness.
Assurance of salvation is also important in Pietism (including forms of Meth-
odism, Lutheranism and Quakers) and Pentecostalism. However, these religions
posit an inward or personal assurance. In Pietism, the doctrine of perfection, or
continuing to mature in faith after receiving salvific assurance, interprets good
works as a spiritual sign and part of the process of perfecting one's faith. Such
2 For a detailed discussion of the relationship between religious beliefs and economic incentives, see
McCleary (2006).
assessments that we have received. One is that, if Weber were alive today and had
access to modern data and statistical tools, he would be carrying out the type of
cross-country empirical work that we have been pursuing. The other is that Weber
thought that religion-notably the contrast between Protestantism and Catholi-
cism-was important for economic development only at a particular stage of history
involving the assimilation of workers into the factory system. He thought that the
religious underpinnings of individual productivity were replaced later by secular insti-
tutions and, hence, that differences in religion no longer mattered much for economic
outcomes at that stage. Thus, Weber did not view religious differences as central in
nineteenth century Europe, and he presumably would not have expected to find
important economic effects of religion in the twentieth century. In this view, Weber
would not find so interesting our quantitative work on modern international data.
3 We have a rough breakdown from other sources into Sunni, Shiite, and other types but do not use this
breakdown in the present analysis. For a discussion in the context of state religions, see Barro and
McCleary (2005b). That research found that overall Muslim adherence, not the breakdown by type,
mattered for the probability of state religion. In contrast, adherence to distinct Christian religions-not
overall Christian adherence-influenced the likelihood of state religion.
4 See Barro and McCleary (2005b) for a fuller discussion of the state-religion variable. This classification
misses the important senses in which, for example, the state religions of England and Iran are not the
same. Fox and Sandler (2003) are assembling a religion and state database in which they classify the
relation between religion and state into four broad groupings: separation of religion and state,
discrimination against minority religions, restrictions on majority religions, and religious legislation.
Although each individual measure is a (0, 1) dummy variable, indexes based on the large number of
separate components would be nearly continuous. Unfortunately, the Fox-Sandler data are available
only since 1990.
Putting the various sources of religion data together, and considering the
availability of data on other variables, we can carry out statistical analysis for up to
81 countries, with a maximum number of country/time observations of 258. The
exact sample size depends on the measure of religiosity. For example, the Gallup
Millennium Survey lacks data on most measures of religious belief, and questions
on personal prayer appear in only two of the World Values Survey (WVS) waves and
the two International Social Survey Program (ISSP) waves.
Until recently, Muslim countries were underrepresented in the surveys. How-
ever, the 2000 World Values Survey wave added a substantial number of predom-
inantly Muslim countries. In terms of most popular religions (not necessarily a
majority of persons adhering to some religion), the full sample has 32 Catholic
countries, 14 Muslim, eleven Orthodox, ten Protestant, six Eastern religions (in-
cluding Buddhist), four other religions (mainly in Africa), two "other Christian"
(one of which is the United States), one Hindu and one Jewish.
The sample has a lot of representation among communist countries-22 that
were communist in 1970, of which only two, China and Vietnam, are still classed as
communist in 2000. Thus, the sample is useful for assessing the effects on religiosity
from current and past communism. (We do not classify communism as itself a form
of religion.) The countries included are richer than the world average, although
poor countries have been increasingly represented over time in the World Values
Survey. For example, the full sample includes seven countries in sub-Saharan
Africa.
5 We adjusted for differences in average levels of responses from the different surveys and time periods
by comparing the overlapping observations for each pair of surveys, for example, 1990 WVS and 1981
WVS. This procedure means that, in some cases, the religiosity questions post-date the growth-rate
observations. However, the instruments used apply to earlier points in time. For further discussion, see
Barro and McCleary (2003).
The first part of Table 1 shows the averages of the religiosity responses for the
countries with data. These averages are unweighted across countries. Weighting by
population would mean that China and India would dominate the world statistics.
This population weighting is appropriate if one wants to consider the position of
the average person in the world. However, to learn about cross-country determi-
nants of religiosity, including influences from governmental institutions, it is more
informative to weight each country roughly the same. The table shows that average
attendance at formal religious services at least weekly is 31 percent, monthly is 41
percent, and participation in personal prayer at least weekly is 57 percent. Average
beliefs were 43 percent for hell, 59 percent for heaven and an afterlife, and 82
percent for the existence of god in some sense. Sixty-nine percent of persons said
that they were at least somewhat religious.
The remainder of Table 1 shows means and standard deviations in 1970 and
2000 for the other variables used in the analysis. Thirty-six percent of the countries
had state religions in 1970, compared to 44 percent in 2000. The increase over time
reflected mainly the 13 previously communist countries, such as Bulgaria and
Ukraine, that implemented state religions between 1990 and 2001.6 Forty-one
percent of countries regulated religion (in the sense described before) in the
1970s. Averages for the religious pluralism indicator (one minus the Herfindahl
index for the ten religion categories noted before) were 33 percent in 1970 and 37
percent in 2000.
The fraction of the overall population designated as nonreligious, according to
Barrett, averaged 11 percent in 1970 and 10 percent in 2000, with the small decline
attributable to the ending of communism in many countries. For 59 never-
communist countries, the average for nonreligion rose from 3 percent in 1970 to
6 percent in 2000. Note again that these averages are unweighted across countries.
In particular, the averages for communist countries do not give a large weight to
the unusually high nonreligion fractions in China, 64 percent in 1970 and 50
percent in 2000.
The bottom part of the table shows the breakdown of religion adherence by
type among persons adhering to some religion; note again that the countries are
equally weighted in these averages. In 2000, the Catholic religion had the largest
share of adherents (36 percent), followed by Muslim (18 percent), Protestant (14
percent), Orthodox (11 percent) and other Christian (9 percent).
Determinants of Religiosity
Table 2 shows estimates of systems in which the dependent variables are survey
responses about monthly participation in formal religious services, weekly personal
prayer, belief in hell and an afterlife, and whether people self-identify as religious.7
These panels combine data from up to seven survey waves. (The Gallup data are
6 For the broader sample of countries considered in Barro and McCleary (2005b), the averages for state
religion were 39 percent in 1970 and 40 percent in 2000, compared to 59 percent in 1900.
7 The dependent variables take the form log[x / (1.02 - x) ], where x is the fraction who attend, believe,
etc. This form constrains the fitted values to lie in the interval (0, 1).
Table 1
Means and Standard Deviations of Variables
These variables are averaged over different surveys, with levels geared to 1990
1970 2000
available only for participation, and the prayer question comes from only four waves.)
The last column, for the fraction of the population designated as having some religious
adherence, is based on Barrett's census/survey information for 1970 and 2000.
The explanatory variables include a single indicator of economic develop-
ment-the log of real per capita GDP- dummy variables for state religion and state
regulation of religion, and dummies for contemporaneous and lagged commu-
nism. We include adherence shares for eight religion groups (where Catholic is the
left-out category, and Buddhist and other Eastern religions were combined due to
limited data), and the measure of religious pluralism (computed from the adher-
ence shares for ten religion groups). The adherence shares are measured relative
to persons identifying with some religion: for example, the proportion of Protestant
adherents out of the population of persons adhering to some religion. The idea is
that the breakdown of adhering persons by type of religion may be exogenous with
respect to measures of religiosity, but the fraction of persons adhering to some
religion (rather than no religion) would surely not be exogenous. Since Catholic is
the left-out category, the coefficients on adherence shares should be interpreted as
effects relative to those for Catholic.8 The samples comprise up to 81 countries and
258 observations for monthly attendance in column 1. Fewer observations are
available for the other systems.
The estimation uses instrumental variables to allow for the potential endogeneity
of per capita GDP with respect to religiosity. We use as instruments two arguably
exogenous determinants of economic development: absolute degrees latitude (which
relates to climate and, hence, to health and agricultural productivity) and a dummy
variable for land-locked status (which affects transport costs)." In effect, the estimation
replaces the log of per capita GDP with the fitted values from a first-stage regression.
This first-stage equation has the log of per capita GDP as the dependent variable and
has as independent variables the two instruments and the other explanatory variables
(which are treated as exogenous). The idea of this instrumental-variable procedure is
to isolate effects of economic development on religiosity, rather than the reverse. The
estimation gives equal weight to each country.
One concern is that explanatory variables other than per capita GDP might
also be endogenous with respect to religiosity. For the measures of religion adher-
ence (among persons adhering to some religion), a possible problem would be
reverse influences of religiosity on conversion into particular faiths. Probably a
more serious concern involves reverse effects of religiosity on the interplay between
state and religion, specifically, on the tendency to have state religion and to
regulate religion.
8 Separate constant terms are included for the different survey sources, but the other coefficients are
constrained to be the same for all surveys. We have tested for whether there are trends in the religiosity
indicators (for given values of per capita GDP and the other explanatory variables). These tests are based
on comparisons across the four WVS waves and between the two ISSP waves. The only statistically
significant trend is for belief in hell, which is increasing over time (holding per capita GDP and other
variables constant).
9 In a very long-term context, land-locked status may be endogenous. Our results are not sensitive to the
exclusion of this instrument.
Table 2
Monthly Consider
attendance Weekly self as Religious
Explanatory at formal personal Belief in Belief in religious fraction of
variable services prayer hell afterlife person population
Table 2-continued
Monthly Consider
attendance Weekly self as Religious
Explanatory at formal personal Belief in Belief in religious fraction of
variable services prayer hell afterlife person population
Number of 22, 37, 21, 31, 15, 29, 21, 34, 15, 26, 34, 15, 21, 39, 14, 62, 80
observations 39, 27, 48, 52 37, 29, 61 37, 29, 61 38, 29, 61
for each 64
equation
R2 for each .80, .57, .61, .71, .66, .46, .66, .51, .61, .47, .49, .61
equation .68, .73, .58, .65 .63, .63, .68, .36, .52, .44,
.68, .69, .60, .74 .43, .62 .47, .56
.69
country's long-ago history of state religion. We did not find important effects on the
probability of state religion from per capita GDP; the composition of religion adher-
ence; or legal structure, such as the extent of constraints on the chief executive. These
results leave open the possibility that other sources of differences in religiosity could
influence the probability of state religion and of state regulation of religion.
One striking result from Table 2 is that per capita GDP has a significantly negative
effect on all of the religiosity indicators.'0 This finding supports the secularization view
as well as the rational-choice perspective of Azzi and Ehrenberg (1975). An irony in this
finding is that the proponents of secularization have been in retreat over the last couple
decades; for example, Berger (1996) recanted his previous stance.
One observation that boosted the arguments of the nonsecularists is that the rich
United States has maintained high levels of religiosity over time-Table 3 shows that
the United States is a substantial outlier in the systems estimated in Table 2. Another
aspect of the debate is that secularists, such as Hume, were unreasonably extreme,
arguing counterfactually that religion would rapidly disappear as a significant social
force. More accurately, secularization can be seen as a gradual tendency. In this
context, Iannaccone (2003) uses International Social Survey Program retrospective
information to construct time series back to the 1920s on participation in formal
religious services for 30 countries. He observes that a steady pattern of diminished
participation-secularization in this sense-applies only to a few countries, such as
Britain, France and Germany. However, no countries show a pattern of steady increase
in participation, and the overall pattern is reduced participation over time.
To quantify our findings, start at sample means in 2000 (shown in Table 1) and
consider an increase of one standard deviation in the log of per capita GDP (by
0.96). In this case, GDP per capita rises from $7,940 (roughly the position of Russia
in 2000) to $20,700 (roughly the position of Italy). Based on our results in Table 2,
this increase in per capita GDP is estimated to lower monthly attendance at formal
religious services by 0.17 (from 0.41 to 0.24), weekly prayer by 0.23 (from 0.57 to
0.34), and belief in hell by 0.16 (from 0.43 to 0.27).
Another finding from Table 2 is that state religion is positively related to
attendance at formal religious services and with beliefs in hell and an afterlife
(columns 1, 3, 4). These results apply for a given regulatory setup and a given
degree of religious pluralism. Our interpretation is that the subsidy element from
state religion motivates more participation, which, in turn, instills greater beliefs.
State religion is, however, not significantly related to the extent of personal prayer
(column 2). That is, organized religion does not show up as a clear substitute or
complement for individual prayer. Given the positive coefficients for beliefs, it is
surprising that state religion is unrelated to the extent to which people self-identify
10 This finding still applies if we do not use instrumental variables to correct for the potential endoge-
neity of per capita GDP. For example, for monthly church attendance in Table 2, column 1, the
estimated coefficient on the log of per capita GDP becomes -0.720 (s.e. = 0.073). For belief in hell in
column 3, the estimated coefficient on the log of per capita GDP becomes -0.671 (s.e. = 0.096).
Inglehart and Baker (2000) also report negative effects of per capita GDP on religiosity in systems
estimated without instrumental variables.
ior is not captured by the model. Monthly attendance at formal services is 60 percent,
compared to the fitted value of 38 percent-corresponding values are 78 percent and
50 percent for weekly prayer and 75 percent and 46 percent for belief in hell. To
appreciate the extent to which the United States is an outlier, we can ask how much
lower U.S. per capita GDP would have to be to fit with observed religiosity, for example,
with monthly church attendance of 60 percent in 2000. The answer is that per capita
GDP would have to be one-third of its actual value: $11,200 rather than $33,300.
Two other outliers with positive residuals are Singapore (44 percent atten-
dance versus fitted of 17 percent and 79 percent belief in hell versus fitted of 41
percent) and Poland (78 percent attendance versus fitted of 34 percent and 66
percent belief in hell versus fitted of 33 percent). Among former communist
countries, Poland is unusual in being extremely religious during the communist
period and then having mild declines in religiosity after the end of communism.
The only contemporaneous communist countries in Table 3 are China (with the lowest
monthly attendance, 3 percent) and Vietnam (also with low attendance, 13 percent).
Many places in western Europe have low religiosity, and much of this behavior
is explained by the model. Denmark and Finland (and also Sweden, which is not
shown) have less than 15 percent monthly attendance rates. However, Finland is
much higher in religious belief than in attendance-31 percent for belief in hell,
not well explained by the model. This pattern of high believing relative to belong-
ing was observed by Davie (1994) to apply in modern Britain. We see something of
this pattern in Table 3-in the United Kingdom, monthly attendance at services is
20 percent, compared with belief in hell of 36 percent.
At the other end, the most religious country in western Europe is Ireland
(68 percent attendance and 53 percent belief in hell, pretty well explained by the
model). Italy and Spain are less religious than Ireland but more religious than the
United Kingdom, France and Scandinavia.
Predominantly Muslim countries exhibit strikingly high levels of belief in
hell-98 percent in Iran, 99 percent in Indonesia, 99 percent in Pakistan and
94 percent in Turkey. Heavily Muslim Nigeria (about 44 percent of the population)
also had 94 percent belief. However, the Muslim countries report more varied
experiences with respect to participation in formal religious services--only
44 percent in Turkey and 47 percent in Iran but 91 percent in Pakistan.
The varied results on participation in religious services in Muslim countries seem
to reflect differences in interpretations by survey takers and respondents in the World
Values Survey of the term "formal religious services." In some countries, women do not
attend services at mosques. However, we have learned that, in some of the surveys, a
"yes" answer to participation in formal religious services required mosque attendance,
whereas, in others, participation in other types of services also counted as formal
participation. Hence, we get puzzling patterns of differences by gender among ten
predominantly Muslim countries in the 2000 World Values Survey. Five countries
(Algeria,Jordan, Morocco, Turkey and Saudi Arabia) show much greater participation
in formal services by men than women. Another five (Bangladesh, Egypt, Indonesia,
Iran and Pakistan) exhibit similar participation rates by gender. India, with a large
number of Muslims in an absolute sense, is similar to the latter group in showing only
Table 3
" If we do not use instrumental variables to correct for the potential endogeneity of the religiosity
variables, the results are qualitatively similar. However, the magnitudes of the coefficients are smaller
than those in Table 4, column 2: 0.0079 (s.e. = 0.0022) for belief in hell and -0.0065 (s.e. = 0.0020)
for monthly attendance at formal religious services.
Table 4
a Other explanatory variables, described below, were included, but coefficients are not
b Buddist plus other eastern religions.
*p < .05, **p < .01.
Notes: The dependent variables are the growth rates of real per capita GDP over 1965-
1985-95. The explanatory variables not shown are the log of per capita GDP in 1965,
years of male secondary and higher school attainment in 1965, 1975 and 1985; rec
expectancy at age one in 1960, 1970 and 1980; average ratios over each period of inve
the log of the total fertility rate in 1960, 1970 and 1980; average ratios for each period o
imports to GDP, filtered for the usual relation of this ratio to the logs of population and a
rate of the terms of trade over each period, interacted with the average ratio of exports
GDP; the average of the Political Risk Services indicator of the rule of law (the value f
appears in the first two equations); and the average for each period of the Freedom H
democracy (electoral rights) and its square. Columns 2 and 3 include the adherence sh
(among persons adhering to some religion) for the eight religion groups shown. The C
the omitted category. The samples for all three columns are the same. Separate constan
for each period. For data sources, see the text.
The coefficients shown in each cell come from joint estimation that pools all of the d
growth from different periods (for varying numbers of countries). Countries are w
irrespective of size. Error terms for a given country from different periods are allowed
Each system treats the two religiosity variables-monthly attendance at formal religio
belief in hell-as endogenous. The instruments are dummy variables for the presence of
and state regulation of religion, the eight religion adherence shares for 1970, and
indicator for religious adherence in 1970.
column 1 are smaller in magnitude and less statistically significant than their
counterparts in column 2. In either case, our central finding is that higher believing
relative to belonging encourages economic growth. To put it another way, growth
is enhanced when the religion sector is unusually productive in the sense that
output (belief related to an afterlife) is high compared to input (attendance).
Given beliefs, more time and resources spent on formal religion can be viewed as
a drain on resources, which detracts from market output (GDP).
The results do not imply that participation in formal religious services is
necessarily negative in a full sense for economic growth. This relation depends on
the extent to which greater attendance at services instills higher beliefs, that is, on
a "religion production function," which we have not yet estimated. However, if we
consider overall differences in religiosity-cross-country variations in belief and
attendance when we assume the typical positive relation between these two highly
correlated variables-the relation with growth turns out to be weak.
The results accord with Weber's emphasis on religion as an influence on
beliefs and, thereby, on individual traits and values. In particular, religion does not
seem to operate as a social organization that enhances productive social capital and
networking. In that scenario, we would anticipate a positive relation between
growth and participation in formal services, rather than the negative relation found
in the data. Thus, the special aspect of religion is belief formation.
In Table 4, column 2, the eight religion adherence shares are jointly highly
significant-the p-value for joint significance is 0.004. Recall that each individual
coefficient should be interpreted as relative to the left-out category of Catholic
adherence. Among the individual effects, the most striking result is the significantly
negative coefficient on Muslim adherence. The coefficient on Protestant adher-
ence is also negative and marginally significant-perhaps Weber would be sur-
prised. None of the other coefficients are individually significantly different from
zero.
12 We also considered the Fox-Sandler indicators for religious mandates on business closings and for
mandatory religious education in public schools. These variables lack any explanatory power for
economic growth.
work ethic = 1.20 + 0.200 (belief in hell) - 0.091 (log per capita GDP)
(0.28 (0.098) (0.027)
This regression suggests that greater belief in hell instills (or goes along with)
stronger work ethic. (The results are a little weaker if belief in hell is replaced by
belief in heaven or an afterlife.) Also interesting is that work ethic declines
significantly with the log of per capita GDP but is significantly higher (for given per
capita GDP and religiosity) in former communist countries. The results are much
weaker for the other two traits considered. For trust, the only statistically significant
coefficient is a positive one for the log of per capita GDP, and the R2 is only 0.21.
For thrift, the only significant coefficient is a positive one for ex-communism, and
the R2 is only 0.13. Thus, Weber may have been right in emphasizing the religion
link with work ethic.
Concluding Observations
References
Arrufiada, Benito. 2004. "Catholic Confession Barro, Robert J. and Rachel M. McCleary.
2003. "Religion and Economic Growth." Ameri-
of Sins as Third-Party Moral Enforcement." Un-
published, Universitat Pompeu Fabra, January.can Sociological Review. October, 68, pp. 760-781.
Azzi, Corry and Ronald Ehrenberg. 1975.Barro, Robert J. and Rachel M. McCleary.
"Household Allocation of Time and Church At-
2005a. "Religion and Political Economy in an
tendance." Journal of Political Economy. February,International Panel, in "Journal for the Scientific
83, pp. 27-56. Study of Religion." Forthcoming.
Barrett, David B. 1982. World Christian Encyclo- Barro, Robert J. and Rachel M. McCleary
pedia, 1st edition. Oxford: Oxford University Press. 2005b. "Which Countries Have State Religions?"
Barrett, David B., George T. Kurian and Todd Quarterly Journal of Economics. November, 120:4,
M. Johnson. 2001. World Christian Encyclopedia, pp. 1331-70.
2nd edition. Oxford: Oxford University Press. Barro, Robert J. and Xavier Sala-i-Martin.
Barro, Robert J. 1997. Determinants of Economic 2004. Economic Growth, 2nd ed. Cambridge, MA:
Growth: A Cross-Country Empirical Study. Cam- MIT Press.
bridge, MA: MIT Press. Berger, Peter L. 1967. The Sacred Canopy: Ele-
Barro, Robert J. and Jong Wha Lee. 2001. ments of a Sociological Theory of Religion. Garden
"International Data on Educational Attainment: City, NJ: Doubleday.
Updates and Implications." Oxford Economic Pa-Berger, Peter L. 1996. "Secularism in Retreat."
pers. July, 53:3, pp, 541-563. National Interest. Winter, 46, pp. 3-12.
Calvin, John. (1584). A Harmonie upon the Smith and the Economics of Religion." Rational-
Three Evangelists, Matthew, Mark and Luke. Lon- ity and Society. April, 3, pp. 156-177.
don: George Bishop. Iannaccone, Laurence. 2003. "Looking Back-
Calvin, John. 1585. An Abridgement of the Insti- ward: A Cross-National Study of Religious
tution of Christian Religion. Edinburgh: Thomas Trends." Unpublished working paper, George
Vautrollier. Mason University, July.
Chaves, Mark and David E. Cann. 1992. "Reg- lannaccone, Laurence R. and Rodney Stark.
ulation, Pluralism, and Religious Market Struc-1994. "A Supply-Side Reinterpretation of the
ture." Rationality and Society. July, 4, pp. 272-290. 'Secularization' of Europe." Journal for the Scien-
Conze, Edward. 1963. "Buddhist Saviors," in tific Study of Religion. March, 33, pp. 76-88.
The Saviour God: Comparative Studies in the Concept Inglehart, Ronald and Wayne E. Baker. 2000.
of Salvation. S.G.F. Brandon, ed. Manchester"Modernization, Cultural Change, and the Per-
U.K.: Manchester University Press. sistence of Traditional Values." American Sociolog-
Davie, Grace. 1994. Religion in Britain sinceical Review. February, 65, 19-51.
1945: Believing without Belonging. Oxford: Keyes, Charles. 1983. "Merit-Transference in
Blackwell. the Kammic Theory of Popular Theravada Budd-
Finke, Roger and Laurence R. Iannaccone.hism," in Karma: An Anthropological Inquiry. C. F.
1993. "Supply-Side Explanations for ReligiousKeyes and E. V. Daniel, eds. Berkeley: University
Change." The Annals of the American Academy ofof California Press, pp. 261-86.
Political and Social Sciences. May, 527, pp. 27-39. Kuran, Timur. 2004. "Why the Middle East Is
Finke, Roger and Rodney Stark. 1992. The Economically Underdeveloped." Journal of Eco-
Churching of America 1776-1990. New Brunswick,nomic Perspectives. Summer, 18, pp. 71-90.
NJ: Rutgers University Press. La Porta, R., F. Lopez-de-Silanes, A. Shleifer
Fox, Jonathan and Shmuel Sandler. 2003. and R. Vishny. 1999. "The Quality of Govern-
"Separation of Religion and State in the 21"tment," Journal of Law, Economics, and Organiza-
Century: Comparing the Middle East and West-tion, 15, April, 222-79.
ern Democracies," presented at the Interna- Marx, Karl. (1859 [1913]). A Contribution to the
tional Studies Association conference, Portland Critique of Political Economy. Chicago: Kerr.
OR, February. McCleary, Rachel M. (2006). "Salvation, Dam-
Freud, Sigmund. 1927. The Future of an Illusion. nation, and Economic Incentives." Journal of
New York: Norton. Contemporary Religion. October, 1.
Gill, Anthony and Erik Lundsgaarde. 2004. Scheve, Kenneth and David Stasavage. (2005).
"State Welfare Spending and Religiosity: "Religion
A and Preferences for Social Insurance."
Cross- National Analysis." Unpublished, Univer-
Unpublished, University of Michigan, March.
sity of Washington, January. Smith, Adam. 1791. An Inquiry into the Nature
Glaeser, Edward E., David I. Laibson, Joseand Causes of the Wealth of Nations, 6th edition.
Scheinkman, and Christine L. Soutter. 2000. London: Strahan.
"Measuring Trust." Quarterly Journal of Economics. Tocqueville, Alexis de. 1835. Democracy in
August, 115, pp. 811-46. America. London: Saunders and Otley.
Heston, Alan, Robert Summers, and Bettina Weber, Max. 1905 [1930]. The Protestant Ethic
Aten. 2002. Penn World Tables Version 6.1. Center and the Spirit of Capitalism, London: Allen & Unwin.
for International Comparisons at the University Wesley, John. 1760. "The Use of Money, a
of Pennsylvania (CICUP), October. Sermon on Luke," in Sermons on Several Occasions.
Hume, David. 1757 [1993]. The Natural History Bristol, J. Grabham and W. Pine, pp. 127-144.
of Religion, edited by J.C.A. Gaskin. Oxford: Ox- Wilson, Bryan. (1966). Religion in Secular Soci-
ford University Press. ety: A Sociological Comment. London: Watts.
Iannaccone, Laurence R. 1991. "The Conse- World Bank. 2005. World Development Indicators
quences of Religious Market Structures: Adam 2005. Washington D.C.: The World Bank.